Hi, Stefano,
2010/10/27 Stefano Babic sbabic@denx.de:
On 10/27/2010 09:40 AM, Jason Liu wrote:
Yes, I haved looked it before and I also tested this patch on MX51babbage 3.0 board. It will printting some confusing information with floating value of MAC address when the board boot up after you set the ethaddr enviorment. The reason is that, when it first call fec_init, the eth address of dev is not intialized and then it using the floating memory value.
Dropping the call is not the solution and works only on the MX51 if the MAC is set into the fuse. Even on the MX51, this is not a general case:
FSL will program all the fuse with MAC address before shipping every chip. This applies to all the i.mx family including imx25/35/51/53/50.
it should be always possible to set the ethaddr via the environment without using the fuse.
Yes, without call this function, the ethaddr evnviroment still working. It will use the MAC set via ethaddr instead of the mac from FUSE. Net framework will do it.
As sure, you should at least see a "Warning : MAC addresses don't match" if the ethaddr is set to a different value as in the fuse.
However, I do not yet understand why the address is not set: fec_probe is called before doing something on the network and at this point the address should be set from fuse.
I have tested it and it's no need to add this function call. You can test it and see the result.
Best regards, Stefano Babic
--
DENX Software Engineering GmbH, MD: Wolfgang Denk & Detlev Zundel HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany Phone: +49-8142-66989-0 Fax: +49-8142-66989-80 Email: office@denx.de =====================================================================